Middle school students present ASEC, an AI system to flag possible school violence

Two Owen Valley Middle School students presented ASEC, an artificial‑intelligence prototype intended to notify staff of possible physical aggression captured on existing school cameras. The project was introduced to the school board by a staff member and was submitted to Purdue University's Presidential AI Challenge.

The students described a two‑model approach: a skeleton‑extraction model that maps body joint relations and an appearance/motion model that analyzes pixel‑level movement over time. "ASEC is a project constructed to examine how artificial intelligence could assist to identify patterns linked to physical aggression," one presenter said, explaining the system flags interactions for staff review rather than making incident determinations.

District staff who brought the students to the meeting said they were impressed with the written proposal and in‑class work that led to the Purdue submission. During questions from board members, the students said their tests produced roughly 97% accuracy. "It reached a 97% accuracy, so I'd say it works pretty good," one presenter said, while also acknowledging the models still require more training and refinement.

Board members asked about practical deployment. The students said the current implementation is a Python (.py) script that runs on a terminal with sample video files; creating a deployable application would require additional engineering, testing and refinement. They also said the team had not filed patents or registrations yet but could pursue legal protections in the future.

The presenters emphasized ethical limits and human oversight: ASEC is intended to highlight potential incidents to alert staff, not to replace human judgment. "We intentionally limited ASEC to assist awareness rather than automate judgment," one presenter said, noting that even low false‑positive rates can have serious consequences in a school setting.

Board members and district staff praised the students and thanked teachers Becky Hayes and Miss Heckman for supporting the project. The board offered good wishes as the students await competition results and discussed possible next steps if the project moves toward wider testing or deployment.

The presentation finished without any board direction to procure or pilot the system; the students' submission to the competition will be judged by Purdue and, if successful, may move to state, regional and national rounds.